Mention metadata management to a non-technical audience and eyes almost certainly glaze over.
The importance of metadata management to IT is (I hope) obvious. Yet metadata management remains poorly invested and often poorly implemented.
Here are our considerations / tips for planning a successful metadata management implementtion
- Start from the business
Historically, metadata management approaches have been driven out of IT – typically driven by the desire to better understand and document extract, transform, load (ETL) processes for the enterprise data warehouse and other large enterprise systems.
These metadata projects have a technical business focus, and tend to present metadata in a technical way.
No wonder we struggle to get and maintain business interest!
Data-driven businesses see metadata as an enabler for the data-driven business: metadata gives data context, allowing business and technical users to more easily find and understand their data.
Rather than being buried in data flows, a business-oriented approach begins with a business problem.
Want to improve customer experience, for example?
Metadata can help you to assess where customer data is stored, how it is used, what missing information may be required to deliver more sophisticated insights, and much more.
A solid metadata capability allows you to quickly assess the impact of changes to data, making your business more able to adapt systems to meet changing customer demands.
And, of course, metadata provide you with traceability – the ability to understand how and where customer data is stored and used. This is critical to meeting regulatory demands for the protection personal data, sanctions monitoring and much more.
Implementing metadata management in support of business outcomes is more likely to engage and excite budget holders than any technically driven requirement.
- Consider your audience
For metadata management to succeed, your approach must consider multiple stakeholders.
Who will be involved in the definition of metadata, and who will be expected to use the metadata?
To some extent the answers to these questions will vary depending on the type of metadata.
Technical metadata, such as that derived from ETL processes, is typically the responsibility of technical / IT staff.
Business metadata, such as the definitions of business policies or business terms, requires the input of business stakeholders.
Value is created when both business and technical stakeholders can collaborate efficiently to build an end to end picture of the enterprise data landscape.
Some stakeholders may have limited part-time roles. Others may require the ability to deliver a range of metadata and be employed in a full time metadata management capacity.
Others may simply need to be informed when changes are proposed to metadata that affects them, or may need to access the repository simply to confirm a definition or policy.
Each of these stakeholders, and any others you may identify, must be catered for in your approach.
- How will you govern your metadata?
Without governance, metadata are just lists of values.
Governance ensures business and technical accountability for metadata, ensures that the metadata initiatives are aligned to business priorities, and helps to improve the accuracy and value of metadata by involving the correct stakeholders at the right time throughout the collection and approval processes.
If data governance is left as an afterthought, you may struggle to measure progress, may find that your metadata does not reflect the business reality, or that you isolate critical stakeholders, thereby limiting the value of your metadata initiative.
At a bare minimum: Are you able to link metadata to a governance structure and have you considered how your documentation and approval processes will work?
More mature organisations may wish to link metadata to impacted business projects (for example master data management initiatives), to business principles and policies (to track compliance), and to impacted business assets such as systems and processes.
How will you govern and manage your metadata?
- How will you structure metadata management?
One goal of metadata management is to have a single definition for each asset at an enterprise level.
In practice, this goal can be difficult to achieve. Different business areas may have common understandings of the same word – in many cases for legitimate business reasons.
For this reason you may want to structure your metadata organisation in order to support differing local views of metadata. For example, your corporate business may define “Churn” differently to your Retail business.
This may mean, in practice, that the Churn report for Corporate will have a different metric to the Churn report for individuals. What is useful is to understand the different metrics in context, and then have some means to aggregate to get an overall Churn score for the Enterprise (if needed)
How will you support different contexts?
Assuming that you support a hierarchical metadata implementation you will then need to consider how lower level organizations will inherit definitions from higher levels, and how lower level definitions will be amalgamated into the global definition.
It may seem simple to have a single enterprise view but in practice this seldom reflects reality and this over simplification leads to resistance and the failure of metadata to deliver value
- How will you make metadata accessible?
If metadata is hard, or impossible, to access then it will not be used.
Some metadata, obviously, is sensitive and access must be restricted,
But broadly speaking, business value and adoption comes when metadata is incorporated into, and supports, standard business functions.
A business analyst, for example, may need access to metadata to contextualize a business requirement and ensure the correct terminology is being used in a Requirements Specification
A data architect may want to access metadata to understand how a logical data model is instantiated in various systems, in order to understand the impact of a required change.
A manager may want to trace the source, understand the metrics and assess the quality of the underlying data in a report.
Making metadata accessible to these users helps to embed the standards into the corporate psyche, delivering value and driving adoption.
- Rome wasn’t built in a day
Metadata is vast and complex.
In order to achieve results you will need to break the workload down (based on business priorities) and distribute the work load amongst many people.
Making metadata accessible means that knowledge workers will start to identify gaps and discrepancies. How will you work with them to enable these gaps, or differences in usage, to be captured and embedded into your metadata store?
If you can get this right your metadata store will continuously grow and improve, adding additional value and driving additional adoption.
Governance in this context is critical. Not only do you need to consider your approval processes, but you also need to manage the lifecycle of metadata assets – including tracking past and future versions of the truth.
What tips do you have for driving the adoption of metadata?
Wonderful content! I actually added your blog to my favorites list and look forward to get the same quality content every time I visit your blog. Thanks a lot.
Thank you – glad you enjoyed it